Only 1 in 20 treatments tested in animals wins approval for people

Key takeaways

• A 2024 analysis of 122 reviews found that, of therapies first tested in animals more than a decade earlier, about half reached any human study, 40% reached a randomized trial and 5% won regulatory approval².
• Of 1,535 news headlines about Alzheimer’s studies in mice, 69.5% never mentioned mice¹.
• Curcumin acts on cells in lab dishes at concentrations far above what people reached in their blood, even after swallowing 12 grams⁴.
• FDA said in 2025 that more than 90% of drugs that look safe and effective in animals never win its approval, and it began shifting toward human-based lab methods³⁵.
• Treat a lab or animal finding as a reason to run human studies, not as proof that something works for you.

“Scientists reverse memory loss.” Headlines like that appear every year. Often the patients were mice, and the headline never says so.

Two researchers, Marcia Triunfol and Fabio Gouveia, measured how often. They took 623 Alzheimer’s papers from 2018 and 2019 that reported experiments in mice, then gathered 1,535 news headlines written about them¹. Seven in ten of those headlines, 69.5%, left the mice out¹.

The habit starts with scientists. When a paper’s own title skipped the mice, 89.6% of the news headlines did too. Those papers also drew more news stories and nearly twice as many tweets¹. One disclosure: the paper lists Triunfol’s employer as Humane Society International, an animal-protection group, and the authors declared no competing interests¹.

This matters to anyone who reads about natural remedies. Many popular claims for curcumin, resveratrol and some mushroom extracts rest mostly on cell and animal studies⁴¹¹¹². Those studies are valuable. They are also chapter one of a long story, and most such stories end before a human chapter gets written.

We walk through why lab and animal findings so often fail to carry over, and how to read the next breakthrough headline.

Cells and mice are where good ideas get their first test

Research usually moves in stages. First come in vitro studies, Latin for “in glass,” done on cells or proteins in a dish. Next come in vivo studies in living animals, most often mice and rats. Only then do studies in people begin.

The early stages are cheaper, faster and less risky. They let scientists see how something might work, look for toxic effects and choose doses before anyone swallows a pill.

Some animal findings do carry over, and there are clues to which ones. A 2006 analysis found that animal studies showing a dose-response pattern, where more of a substance produced more effect, were more likely to hold up later in people⁶.

For decades, U.S. drug development leaned on animal testing before human trials. Then, in late 2022, Congress passed the FDA Modernization Act 2.0, which allowed alternatives such as cell-based tests and computer models to support requests to begin human trials³.

Most animal successes never reach a pharmacy

Two analyses, 18 years apart, tell a consistent story.

In 2006, researchers David Hackam and Donald Redelmeier tracked 76 animal studies published in top journals such as Science, Nature and Cell, each cited more than 500 times⁶. What became of them:

• 37% were later confirmed in randomized human trials.
• 18% were contradicted by human trials.
• 45% were never tested in people at all.

Confirmation took a median of seven years, and eight of the treatments were approved for patients⁶. Few of the animal studies assigned animals to groups at random or kept the people measuring results unaware of which animals got the treatment⁶.

In 2024, a team led by Benjamin Ineichen pooled 122 systematic reviews covering 4,443 animal studies, 367 therapies and 54 diseases². Among therapies first tested in animals more than 10 years earlier, about half reached any human study, 40% reached a randomized trial and 5% won regulatory approval². That 5% is the “1 in 20” in our headline.

Counting every therapy, including newer ones, the shares were a bit lower: 45%, 36% and 4%². For the few that won approval, it took a median of 10 years².

This is the surprise. For 62 therapies studied heavily in both animals and people, the authors found that animal and human results broadly agreed². About 79% of those animal studies were positive, compared with 61% of human studies and 50% of randomized trials².

So animal studies don’t simply mislead. The steepest drop-off comes later, as treatments face larger and stricter human trials.

The authors’ leading explanation is weak design, such as skipping randomization or blinding, in both animal and early human studies². Mismatches matter, too. Stroke treatments, for example, were often tested in young, healthy animals, while stroke patients tend to be older and have other illnesses².

FDA’s own 2025 roadmap is blunter: more than 90% of drugs that look safe and effective in animals never win FDA approval, mostly because of safety or effectiveness problems in people³.

Mice are not small people

Inflammation research shows why. For years, drug makers tried to calm the runaway inflammation that can kill patients in intensive care. A 2013 paper in the Proceedings of the National Academy of Sciences counted nearly 150 such trials, with no successes⁷.

The same paper, from a large research team, offered a clue. The researchers tracked which genes switched on or off after burns, trauma or a dose of bacterial toxin, in patients and in matching mouse models⁷. Burned patients and injured patients showed nearly the same gene pattern, a 0.91 match on a scale where 1 is identical. Mice and people matched at 0.09 or lower⁷.

The clocks ran differently, too. In the mice, the gene upheaval faded in a matter of hours or days. In patients, it lingered for one to six months or longer⁷.

Not everyone agreed. In 2014, two other researchers reanalyzed the same data, looking only at genes that changed in both species, and argued that mice mimic human inflammation well. A coauthor of the original study called that cherry-picking⁸. The debate is a lesson in itself: how alike mice and people look depends on the question you ask.

Safety can split between species as well. FDA notes that an experimental antibody drug called TGN1412 seemed safe in monkeys but caused a life-threatening immune reaction in human volunteers³.

What works in a dish may never reach your blood

Curcumin, the best-known compound in turmeric, shows how test-tube results can mislead. In lab studies it acts on a long list of cell targets, usually at concentrations in the micromolar range. In one example, it harmed human kidney cells at 15.2 micromolar⁴.

Reaching those levels in a person is another matter. In one study, people swallowed single doses of 10 to 12 grams of curcumin, and it turned up in the blood of only two of six people⁴.

The highest level, in one person given 12 grams, was 156 nanomolar⁴. That’s about one-hundredth of the lab concentration above.

Curcumin is also fragile. In lab solutions at body temperature, half of it breaks down in under 10 minutes⁴. A 2017 review by medicinal chemists in the Journal of Medicinal Chemistry argued that curcumin interferes with so many lab tests that it often produces false signals of activity⁴.

By 2017, more than 120 clinical trials had tested curcumin and related compounds. The authors wrote that none of the double-blind, placebo-controlled trials had succeeded⁴.

The human picture has moved a little since then. NCCIH, the federal center for complementary health research, says in its turmeric fact sheet, last updated in April 2025, that pooled studies show positive early evidence for knee osteoarthritis pain and stiffness, though better studies are needed¹⁰. For other uses, it says the evidence is insufficient¹⁰.

Attempts to fix absorption carry their own risks. Liver injury has been reported with some products designed for better absorption¹⁰. If you notice tiredness, nausea, poor appetite, dark urine or yellowing skin or eyes, stop taking it and talk with your health care provider¹⁰. Turmeric supplements may be unsafe during pregnancy¹⁰.

Animal doses don’t convert one-to-one, either. FDA’s dosing guidance estimates a human equivalent by body surface area, dividing a mouse dose in milligrams per kilogram by 12.3⁹. A headline saying “high doses helped mice” tells you little about what a person would need or could tolerate.

Resveratrol, the compound linked to red wine, followed a similar path. Studies in the early 2000s reported longer lives in yeast, worms, flies and rodents. But the body breaks resveratrol down quickly, and a 2014 editorial found that human trials had produced conflicting results¹¹.

Some lab leads do reach people, and mushrooms show both paths

Medicinal mushrooms show what climbing the ladder looks like. Turkey tail mushroom contains a compound called PSK. Lab and animal studies looked at its effects on immune cells, and then researchers tested it in people with cancer, alongside chemotherapy¹².

In stomach cancer, those human studies suggest better survival when PSK is added¹². Japan has approved PSK as a cancer treatment¹².

Reishi, another popular mushroom, has a thinner human record made up of small studies, including a 36-patient lung cancer trial¹². In the United States, neither mushroom is FDA-approved for any condition. Both are sold as supplements, so products can vary¹². If you have cancer, talk with your cancer care team before adding any mushroom product.

Washington is moving away from animal testing

On April 10, 2025, FDA announced a plan to reduce, refine or potentially replace animal testing in drug development, starting with monoclonal antibodies, a type of lab-made antibody drug⁵. In their place, it encourages computer models, lab-grown mini-organs called organoids, and organ-on-a-chip devices built to mimic human organs⁵. The stated goal is to make animal studies “the exception rather than the norm” in early safety testing within three to five years³.

On December 2, 2025, FDA issued draft guidance saying six-month monkey toxicity studies can be cut or reduced for certain antibody drugs. A typical antibody drug program can use more than 100 monkeys at about $50,000 each¹³. FDA’s announcement presented it as a draft that would take effect only once finalized¹³. Check FDA’s website for its current status.

NIH followed in July 2025. According to an NIH notice dated July 18, 2025, the agency will no longer create new funding opportunities focused only on animal models of human disease, though it still funds animal studies that are scientifically justified¹⁴. As of mid-July 2025, NIH had not published a final written policy, and some scientists pushed back¹⁵. One University of California, San Francisco neuroscientist warned that a rapid phase-out would be a disaster for brain research, which often depends on studying whole animals¹⁵.

Many natural-health readers care about animal welfare and about evidence, and this shift touches both. Human-based tools may predict some harms better.

In one validation study, a human liver-chip correctly flagged 87% of drugs known to have injured patients’ livers³. But a chip isn’t a person, either. New methods still have to prove they predict human results, and none replaces a well-run human trial.

The evidence ladder, from lab dish to your kitchen table

Use this ladder to place any claim you read. Our scale treats a claim resting only on lab and animal work as not enough evidence for people, and one with mostly lab and animal work plus early human data as promising but early.

Rung	What it is	What it can show	What it can’t show	Typical rating for a health claim
Lab dish	Cells or proteins in a dish	Whether a compound does anything, and how	Whether it reaches your tissues at that level	Not enough evidence
Animal study	Mice, rats, sometimes monkeys	Effects in a living body, early safety signals	Whether people respond the same way	Not enough evidence
Small human study	Pilot trials, often without a placebo group	Whether people tolerate it; early hints	Reliable proof of benefit	Promising but early
Randomized trial	Chance decides who gets the treatment or a placebo	Whether it works, and how often it harms	Results in groups that weren’t studied	Strong evidence, if results repeat
Systematic review	All the trials weighed together	The overall answer and how certain it is	Anything the trials inside got wrong	Strong evidence, if the trials are sound

How popular lab-born claims rate today

Claim	What the evidence shows	Our rating
Turmeric or curcumin eases knee osteoarthritis pain	Pooled studies positive; better studies needed10	Promising but early
Curcumin treats or prevents other conditions, such as cancer	Mostly lab and animal work; NCCIH calls the evidence insufficient410	Not enough evidence
Resveratrol supplements extend human life	Longer lives in some lab animals; human results conflict11	Not enough evidence
PSK from turkey tail, added to chemotherapy for stomach cancer	Human studies suggest better survival; approved in Japan, not by FDA12	Promising but early
Reishi mushroom treats cancer	Small human studies only12	Not enough evidence
Anti-inflammatory drugs that worked in mouse models save critically ill patients	Nearly 150 trials failed7	Evidence says no
A positive animal study predicts an approved human treatment	About 5% of animal-tested therapies won approval2	Evidence says no, as a general rule

Questions readers ask

The bottom line

Lab and animal studies are where nearly every remedy begins, natural or pharmaceutical. They point to promising ideas and screen out dangerous ones. But only a small share of treatments that shine in a dish or a mouse go on to help people.

When you see “breakthrough,” look for the species, the dose and the next rung of evidence. A finding in mice is a reason for hope and for human research. It is not yet a reason to buy.

Related reading

• Most health products aren’t scams, but many just haven’t been proven
• What “cruelty-free” on a label does not promise
• Plants gave us Taxol and digoxin, yet the herb isn’t the drug
• How systematic reviews work, and why cranberries failed one and passed the next
• How clinical trials decide, when most treatments tested in people fail

Sources

1. Triunfol M, Gouveia FC. “What’s not in the news headlines or titles of Alzheimer disease articles? #InMice.” PLOS Biology, 2021. journals.plos.org
2. Ineichen BV, Furrer E, Grüninger SL, Zürrer WE, Macleod MR. “Analysis of animal-to-human translation shows that only 5% of animal-tested therapeutic interventions obtain regulatory approval for human applications.” PLOS Biology, 2024. pmc.ncbi.nlm.nih.gov
3. U.S. Food and Drug Administration. “Roadmap to Reducing Animal Testing in Preclinical Safety Studies.” 2025. fda.gov
4. Nelson KM, Dahlin JL, Bisson J, Graham J, Pauli GF, Walters MA. “The Essential Medicinal Chemistry of Curcumin.” Journal of Medicinal Chemistry, 2017. pmc.ncbi.nlm.nih.gov
5. U.S. Food and Drug Administration. “FDA Announces Plan to Phase Out Animal Testing Requirement for Monoclonal Antibodies and Other Drugs.” April 10, 2025. fda.gov
6. Hackam DG, Redelmeier DA. “Translation of research evidence from animals to humans.” JAMA, 2006. deepdyve.com
7. Seok J, Warren HS, Cuenca AG, et al. “Genomic responses in mouse models poorly mimic human inflammatory diseases.” Proceedings of the National Academy of Sciences, 2013. pmc.ncbi.nlm.nih.gov
8. Science News. “Debate rages over mouse studies’ relevance to humans.” August 5, 2014. sciencenews.org
9. U.S. Food and Drug Administration. “Guidance for Industry: Estimating the Maximum Safe Starting Dose in Initial Clinical Trials for Therapeutics in Adult Healthy Volunteers.” July 2005. fda.gov
10. National Center for Complementary and Integrative Health. “Turmeric.” NIH, updated April 2025. nccih.nih.gov
11. Ponzo V, Soldati L, Bo S. “Resveratrol: a supplementation for men or for mice?” Journal of Translational Medicine, 2014. translational-medicine.biomedcentral.com
12. National Cancer Institute. “Medicinal Mushrooms (PDQ), Patient Version.” Updated July 11, 2024. cancer.gov
13. U.S. Food and Drug Administration. “FDA Releases Draft Guidance on Reducing Testing on Non-Human Primates for Monoclonal Antibodies.” December 2, 2025. fda.gov
14. National Institutes of Health, Office of Extramural Research. “How Does the NIH Initiative to Prioritize Human-Based Research Affect Research Proposing the Use of Laboratory Animals?” July 18, 2025. grants.nih.gov
15. The Transmitter. “NIH proposal sows concerns over future of animal research.” July 15, 2025. thetransmitter.org